- Record: found
- Abstract: found
- Article: found
A human expression system based on HEK293 for the stable production of recombinant erythropoietin
research-article
Author(s):
Christine Lin Chin
1 ,
Justin Bryan Goh
1 ,
Harini Srinivasan
2
,
3 ,
Kaiwen Ivy Liu
3 ,
Ali Gowher
3 ,
Raghuvaran Shanmugam
3 ,
Hsueh Lee Lim
1 ,
Matthew Choo
1 ,
Wen Qin Tang
1 ,
Andy Hee-Meng Tan
1 ,
Terry Nguyen-Khuong
1 ,
Meng How Tan
2
,
3
,
,
Say Kong Ng
1
,
Publication date (Electronic):
14 November 2019
Journal:
Scientific Reports
Publisher:
Nature Publishing Group UK
Read this article at
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
Mammalian host cell lines are the preferred expression systems for the manufacture of complex therapeutics and recombinant proteins. However, the most utilized mammalian host systems, namely Chinese hamster ovary (CHO), Sp2/0 and NS0 mouse myeloma cells, can produce glycoproteins with non-human glycans that may potentially illicit immunogenic responses. Hence, we developed a fully human expression system based on HEK293 cells for the stable and high titer production of recombinant proteins by first knocking out GLUL (encoding glutamine synthetase) using CRISPR-Cas9 system. Expression vectors using human GLUL as selection marker were then generated, with recombinant human erythropoietin (EPO) as our model protein. Selection was performed using methionine sulfoximine (MSX) to select for high EPO expression cells. EPO production of up to 92700 U/mL of EPO as analyzed by ELISA or 696 mg/L by densitometry was demonstrated in a 2 L stirred-tank fed batch bioreactor. Mass spectrometry analysis revealed that N-glycosylation of the produced EPO was similar to endogenous human proteins and non-human glycan epitopes were not detected. Collectively, our results highlight the use of a human cellular expression system for the high titer and xenogeneic-free production of EPO and possibly other complex recombinant proteins.
Related collections
Most cited references45
- Record: found
- Abstract: not found
- Article: not found
Biopharmaceutical benchmarks 2014.
Gary F Walsh (2014)
- Record: found
- Abstract: found
- Article: found
Genome dynamics of the human embryonic kidney 293 lineage in response to cell biology manipulations
Yao-Cheng Lin, Morgane Boone, Leander Meuris … (2014)
The human embryonic kidney (HEK) 293 cell line and its derivatives are used in experiments ranging from signal transduction and protein interaction studies over viral packaging to rapid small-scale protein expression and biopharmaceutical production. The original 293 cells1 2 3 were derived in 1973 from the kidney of an aborted human embryo of unknown parenthood by transformation with sheared Adenovirus 5 DNA. The human embryonic kidney cells at first seemed recalcitrant to transformation. After many attempts, cell growth took off only several months after the isolation of a single transformed clone. This cell line is known as HEK293 or 293 cells (ATCC accession number CRL-1573). A 4-kbp adenoviral genome fragment is known to have integrated in chromosome 19 (ref. 4) and encodes for the E1A/E1B proteins, which interfere with the cell cycle control pathways and counteract apoptosis5 6. Cytogenetic analysis established that the 293 line is pseudotriploid7. Given the broad use of 293 cells for biomedical research and virus/protein production, we decided to perform a comprehensive genomic characterization of the 293 cell line and the most commonly used derived lines (Fig. 1a) to better understand the dynamics of the 293 genome under the procedures commonly used in biotechnological engineering of mammalian cell lines. First among these derived lines, we analysed 293T, which expresses a temperature-sensitive allele of the SV40 T antigen8 9. This enables the amplification of vectors containing the SV40 ori and thus considerably increases the expression levels obtained with transient transfection. SV40 T forms a complex with and inhibits p53, possibly further compromising genome integrity10. The original 293 line was suspension growth-adapted through serial passaging in Joklik’s modified minimal Eagle’s medium11. Full adaptation took about 7 months, and the first passages were so difficult that the few cells that grew through are likely to have been almost clonal (Dr Bruce Stillman, personal communication). The fully adapted cell line is known as 293S and is also analysed here. Subsequently, this line was mutagenized with ethylmethanesulfonate (EMS) and a Ricin toxin-resistant clone was selected out. The line lacked N-acetylglucosaminyltransferase I activity (encoded by the MGAT1 gene) and accordingly predominantly modifies glycoproteins with the Man5GlcNAc2 N-glycan. Then, a stable tetR repressor–expressing clone of this glyco-engineered cell line was derived to enable tetracyclin-inducible protein expression12. This cell line is widely used for the production of homogenously N-glycosylated proteins and will be referred to as 293SG. Apart from these four cell lines in common use, we also analysed the genome of two 293-derived lines used in our laboratory for protein–protein interaction screening (293FTM) and glyco-engineering (293SGGD; details in Supplementary Information). In our study, following genomic studies of other human cell lines13 14 15, we aim to provide a full-genome resource for these cell biology ‘workhorse’ cell lines while developing the necessary tools to make such resources easily available. This enables all researchers using the 293 cell lines to make fully informed analyses of genomic regions of interest to their studies, without expert bioinformatics skills. We also map the genomic changes accumulating after standard laboratory cell culturing (passaging and freezing), providing a way to assess genomic stability of each line. Furthermore, we present a workflow for determining the insertion sites of viral sequences and plasmids based on the genome sequencing data. The extreme chromosome structure diversity/plasticity in the 293 cell line underlies a novel application: selection of 293 clones surviving stringent selective conditions (in our case: ricin toxin), followed by whole-genome analysis of copy number alterations, can effectively pinpoint the genomic region(s) that contain the gene(s) that is required for adaptation to those selective conditions. Results 293 cell lineage genome, karyotype and transcriptome For genome resequencing, we used complete genomics (CG) high-coverage genome sequencing technology16 (Supplementary Methods; data set summary in Supplementary Tables 1 and 2, and sequencing quality overview in Supplementary Fig. 1). 293 cells are of female provenance, as we find no trace of Y-chromosome-derived sequence in our data sets. The mitochondrial sequence belongs to the oldest European haplogroup U5a1 (refs 17, 18). Furthermore, we applied multiplex fluorescence in situ hybridization analysis to our 293 lines (Supplementary Data 1). A wide diversity of karyotypes was found, also within each clone, with some chromosomal alterations relative to the human reference genome present in almost all cells, and others in only a small proportion of cells. Overall, the pseudotriploidy of the 293 lineage was confirmed both by CG sequencing and karyotyping. To further define the 293 cell lineage and to enable the future development of cell line authentication genotyping assays, we analysed which single-nucleotide polymorphisms (SNPs) in protein-coding regions were common to the six sequenced 293 cell lines (Supplementary Data 2) and we manually curated the functional annotation of all novel (that is, not present in dbSNP) 293-defining SNPs (Supplementary Data 2). The genome-wide 2-kb-resolution sequencing coverage depth analysis provides a 2-kb-window copy number that is relative to the genome-averaged copy number in that particular genome. To obtain the absolute copy number, an independent data source is required. For this purpose, we used the Illumina SNP-array-determined genome-averaged ploidy number. The resulting calibrated 2-kb-resolution copy number shows very good consistency with the lower-resolution Illumina SNP-array copy number variant (CNV) results (Spearman rho=0.67–0.80, depending on the cell line; P 300,000 Broad Institute mouse/human genome-wide shRNAs mapped uniquely to the human RefSeq gene collection, visualized these in an IGV annotation track (Fig. 6b) and investigated which of these targets are mutated in our 293 cell lines. Depending on the cell line, this was the case for 9,608–11,534 (~6% of the ones that aligned) of these shRNAs, which may render these nonfunctional in gene silencing. The 293 line was also one of the many cell lines selected for analysis by the ENCODE project37. Several data sets that are highly complementary to ours and deal, for example, with epigenomics are becoming available in this way. We will be updating our web interfaces for the 293 genome with these and other generated data sets on an ongoing basis. Discussion Cell lines are instrumental for our growing understanding of mammalian biology and for biopharmaceutical production. 293 cells are second only to HeLa cells in the frequency of their use in cell biology (a search in PubMed for this cell line and its most popular derivatives yields ~20,000 hits). They are second only to CHO cells for their use in biopharmaceutical production (and take the prime spot for use in small-scale protein production and in viral vector propagation). However, 293 cells were at some point derived from an individual human embryo with a genome different from the reference. Moreover, the establishment of the cell line and its continuous growth in vitro impose selective conditions on the cells, which are often adapted to through mutation. Thus, the human reference genome sequence provides only a partial understanding of the genome of human cell lines. As genome-wide short interfering RNA resources are now available for human cells38 39, and as sequence-specific genome-engineering tools are rapidly becoming standard tools for mammalian cell genetic engineering40 41 42, a sequence and average copy number level knowledge of the entire genomes of the cell lines under study is of great advantage. Furthermore, the cell-line-specific genome sequences reported here will also be beneficial in the interpretation of RNA-seq and proteomics experiments that make use of these cells. 293 cells have been cultivated for decades in different laboratories, which most likely has led to different progressive genome structure alterations. This may underlie the sometimes different conclusions drawn from experimentation with 293 cell lines (and many other cell lines). All cell lines sequenced here are available to the research community. Up to the level of sensitivity afforded by our sequencing approach (single copy plasmid insertions were easily detected), these cell lines have no inadvertent virus insertions, which should help to put at rest some of the concerns towards the use of the 293 cells for biopharmaceutical production. The analytical tools we provide here for integrated plasmids and viral sequences will be very valuable in fully characterizing cell lines used for the production of biopharmaceuticals, both towards the copy number and stability of the inserted plasmids and the validation that such cell lines are free of inadvertent viral sequence contamination. We have shown that comparative sequencing of several 293 lines of the same descent reveal genomic copy number alterations that explain diverse phenotypes of the lineage and its subclones. Extensive further experimentation is now required to validate the role of these CNVs in cellular transformation, suspension growth adaptation and metabolism. We hope that such studies will contribute to the design of new generations of 293 cells that are even better adapted to experimental and pharmaceutical production requirements, and the knowledge gained may be instructive in how to directly engineer other human cell lines. Furthermore, it is clear from our data that the standard practice of generating a stable clone through transfection and selection will result in the isolation of one geno/karyotype present in the parental cell line. Thus, any phenotype of the resulting stable transfectant may be because of the integrated transgene, or may be because of a genomic difference between the new line and its parental line. Consequently, such experiments should be interpreted with great caution and these data argue for the use of efficient transient transfection or propagation of a polyclonal pool of stable transfectants (in which case a more representative population of the parental cells is analysed) in, for example, quantitative signal transduction studies that use 293 cells (as is used in many drug screening and ‘omics’ experiments). However, the other side of the medal is that there is promise in a potential forward genetics approach offered by analysing phenotype-causative focal copy number variations (in particular full deletions) in 293-derived clones selected for adaptation to new growth conditions (such as high-cell density cultivation while producing biopharmaceuticals, virus infection, activation of particular signal transduction pathways and so on). This approach is made possible by the apparent property of 293 cells to have lost control over chromosomal structure to a great extent. Consequently, a culture of 293 cells should be considered as an entire 'population' of individual cells with different chromosomal structure makeup. Copy number variations are easy to identify at high resolution using high-coverage resequencing. Further experimentation will reveal whether phenotype-selected copy number variations can always be distinguished from such variations that occur randomly. In this perspective, genomic diversity of the 293 cell line might prove to be an experimental opportunity and might further enhance its role as a provider of knowledge on human cell biology. Methods Cell cultivation for DNA and RNA preparation All cell lines were cultured from frozen stocks at 37 °C in Dulbecco’s Modified Eagle Medium (DMEM; Invitrogen) supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, 100 U ml−1 penicillin G, 110 mg l−1 sodium pyruvate and 100 μg ml−1 streptomycin. All lines were routinely split twice a week, when ~80% confluency was reached. Depending on the cell line, the dilution was between 1:3 (293A) and 1:20 (293T). To prepare genomic DNA, ~30 million cells were harvested for each line. The genomic DNA was extracted and purified using the Gentra Puregene Cell kit (Qiagen GmbH, Hilden, Germany) with RNAse treatment of the samples, according to the manufacturer’s instructions. DNA concentrations were determined fluorimetrically with the Quant-iT PicoGreen dsDNA Reagent (Molecular Probes, Life Technologies Ltd., Paisley, UK). For RNA preparation, the cell lines were cultured in 75-cm2 filter cap flasks in a humidified, 8% CO2 atmosphere incubator in DMEM/Ham’s F12 (DMEM/F12; Invitrogen) supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, 100 U ml−1 penicillin G and 100 μg ml−1 streptomycin. Flask positions in the incubator were randomized daily to correct for potential temperature biases. Total RNA was extracted from three replicates of each cell line using Qiagen’s RNeasy Midi kit according to the manufacturer’s instructions, including an on-column DNase-I digest. Concentrations were determined with a NanoDrop ND-1000 spectrophotometer (Thermo Scientific), and RNA quality was assessed on a 2100 Bioanalyzer using RNA 6000 Pico chips (Agilent Technologies). All samples had an RNA integrity number of 9.5 or better. For the RT–qPCR validation of miRNA expression levels, procedures were identical except that the small RNAs were isolated using the miRCURY RNA isolation kit Cell and Plant (Exiqon), again according to the manufacturer’s instructions. Exon arrays After spiking total RNA from each cell line with bacterial poly-A RNA-positive controls (Affymetrix), every sample was reverse-transcribed, converted to double-stranded cDNA, in vitro-transcribed and amplified using the Ambion WT Expression Kit. The obtained single-stranded cDNA was biotinylated after fragmentation with the Affymetrix WT Terminal Labeling kit as outlined in the manufacturer’s instructions. The resulting samples were mixed with hybridization controls (Affymetrix) and hybridized on GeneChip Human Exon 1.0 ST Arrays (Affymetrix). The arrays were stained and washed in a GeneChip Fluidics Station 450 (Affymetrix) and scanned for raw probe signal intensities with the GeneChip Scanner 3000 (Affymetrix). For the processing of the data, see extended experimental procedures. Exon-array data analysis We used a combination of the R Statistical Software Package (www.r-project.org) and Affymetrix Power Tools (APT; Affymetrix) for the quality control and differential expression analysis of the exon-array data, partly as described earlier43. The full R code and APT commands are available as in Supplementary Data 9 and 10). Briefly, exon- and gene-level intensity estimates were generated by background correction, normalization and probe summarization using the robust multi-array average algorithm with APT. At the gene level, after quality control of the raw data in R, genes of which the expression was undetected in all six lines were removed from further analysis, as were the genes of which expression was below the estimated noise level in all lines. This noise level threshold was set at the signal intensity level that eliminated ‘detection’ of expression of more than 95% of the genes on the Y-chromosome, which is absent from the HEK293 lineage (which was derived from a female embryo) and thus serves as an appropriate internal negative control. Differential gene expression analysis was performed for the relevant cell line pairs using a linear model fit implemented in the R Bioconductor package Limma44, considering only core probe sets. The Benjamini–Hochberg (BH) method was applied to correct for multiple testing. Lists of significantly up- and downregulated genes (BH-adjusted P values<0.01) with a minimal twofold change in expression were subjected to functional enrichment analysis using DAVID45 and IPA (Ingenuity Systems, www.ingenuity.com), transcription factor regulation prediction using DiRE46 and manual inspection. Those lists are available as Supplementary Materials. For integration in the IGV genome browser36, we chose to display all genes found to be differentially expressed (BH-adjusted P value<0.01) in the pairwise comparison of interest, irrespective of their log2-fold change, which is displayed as a function of the bar height. The ‘web link to gene expression data’ track links every gene of which expression was detected to a table with the statistical details. The mean exon expression values in the IGV ‘mean probe set expression’ tracks represent the log2 signal values of the filtered extended exon probe sets, that is, after removal of undetected, cross-hybridizing and noisy probes. CG sequencing and analysis Anticipating the pseudotriploidy of the HEK293 genome, genomic DNA from each cell line was submitted to CG’s sequencing service16 (detailed in Supplementary Information) with the request to maximize the sequencing machine’s output to achieve the highest coverage possible, yielding 158~287 Gb of mapped reads of which 122~190 Gb of reads mapped with an expected paired distance (Supplementary Tables 1 and 2). The raw data were analysed with version 1.11 of the company’s analysis software and processed with CGAtools v1.5 (http://cgatools.sourceforge.net/). This pipeline entails read mapping followed by local reassembly of reads that map to a region in which deviation from the reference sequence is suspected from the mapping results. This is then used as the input for SNP and small indel calling. A second analysis focuses on copy number variation (see Supplementary Note 1) and uses the genome-normalized average sequence coverage as input, together with the genome-normalized sequence coverage of 46 normal diploid human genome-resequencing data sets (baseline genome) for the area under analysis. These latter data are used to correct the coverage for sequence-specific biases in the sequencing workflow. The output of this analysis is 2-kbp-resolution copy number expressed as a factor relative to a copy number of 2. As described in the main text, we derived true copy number from these data through calibration with genome-weighted average ploidy as derived from Illumina SNP-array data (Supplementary Table 6). A third analysis uses the paired-end reads of which the mate pairs do not map to a continuous stretch of the human reference genome sequence, and which thus provide evidence for chromosomal rearrangements. These reads are de novo assembled into ‘junction sequence contigs’ that contain the information about the breakpoints involved in such chromosomal rearrangements. The CG raw data and initial analysis results were processed by CGAtools v1.5 (http://cgatools.sourceforge.net/) with scripts from the CG user community tool repository and our in-house scripts (see Supplementary Note 2). To enable independent analysis of the data, we mapped the sequencing reads to the human reference genome, build hg18, using RTG Investigator from Real Time Genomics (http://www.realtimegenomics.com/) with default setting (maximum mate-pair insert size: 1,000, minimum insert size 0 and report the maximum best five matches). Upon mapping, SNP and small indel calling were also performed using the RTG software Investigator. Only SNP/indels passing the quality filter (called in more than half of the reads and covered by less than 200 × coverage to avoid variant calling in highly repetitive regions) were kept for further analysis. The lists of SNPs and indels called either by CG or RTG were merged by vcftools47. To remove platform-specific artifacts from the CG sequencing, the extended variant list was filtered using ANNOVAR48, to remove variants located in a region where less than 30% of the CG69 data sets had sequencing information. We then functionally annotated this filtered extended variant list by ANNOVAR. We used GenomeComb (http://genomecomb.sourceforge.net/) to reformat the SNV calling results from CG for the six cell lines49. In order to increase the number of concordants between cell lines and reduce the false-positive SNV calling rate, we used the obligatory filtering strategy: remove uncertain calls and filtered based on the variant score reported from CG in each cell line. Variant scores lower than the reported average variant score were removed. The SVs detected from CG analysis were first filtered with cgatools against the publicly available Yoruban (NA19238) CG genome data set, to remove frequently occurring SVs. SVs in the 293-derived cell lines were further filtered against the 293 line and we only retained those with low frequency (<10%) in the CG69 population for further manual inspection. SNP-array procedures Genomic DNA (same sample as used for genome sequencing) of each cell line was analysed using the Illumina HumanCytoSNP-12 v2.1 SNP-array, entirely according to the manufacturer’s instructions. For analysis, we used the ASCAT algorithm, which accurately determines allele-specific copy numbers in tumours and aneuploid cell lines by estimating and adjusting for overall ploidy and effective tumour fraction in the sample50. ASCAT uses the raw BAF and logR data of the Illumina HumanCytoSNP-12 v2.1. Author contributions Y.-C.L. designed experiments and analysed the CG data, plasmid insertion site detection and data integration, under the scientific supervision of Y.V.d.P. M.B. carried out exon-array experiments and data analysis, qPCR validation of array data and general data mining. L.M. conducted mitochondrial haplotype study, PCR validation of plasmid insertion sites and general data mining. I.L. performed general data mining, under the scientific supervision of J.T. N.V.R. carried out multiplex fluorescence in situ hybridization data generation and analysis, under the scientific supervision of F.S. A.S.: 293 Variant Viewer website construction. J.R. assisted in GenomeComb analysis of CG data. M.M. carried out SNP arrays, under the scientific supervision of D.L. S.P. helped with general bioinformatics assistance. R.D. and J.C. performed CG data acquisition. N.C. carried out project initiation and design and scientific supervision. N.C., M.B., Y.C.-L. and L.M. co-wrote the manuscript. Additional information Accession codes: Complete Genomics sequencing data have been deposited in the European Nucleotide Archive (ENA) under the accession code PRJEB3209. The Affymetrix exon-array data have been deposited in the ArrayExpress Archive under the accession code E-MEXP-3516. How to cite this article: Lin, Y.-C. et al. Genome dynamics of the human embryonic kidney 293 lineage in response to cell biology manipulations. Nat. Commun. 5:4767 doi: 10.1038/ncomms5767 (2014). Supplementary Material Supplementary Information Supplementary Figures 1-11, Supplementary Tables 1-7, Supplementary Notes 1-5, Supplementary Methods and Supplementary References Supplementary Data 1 M-FISH (Multicolor-FISH) analysis Supplementary Data 2 Manually annotated single-nucleotide polymorphisms (SNPs) and structural variants (SVs) Supplementary Data 3 Copy number variation per chromosome Supplementary Data 4 Approximate copy number of citric acid cycle-related genes in the 6 sequenced cell lines Supplementary Data 5 Gene-level differential expression Supplementary Data 6 Copy number variation, minor allele frequency, ploidy level determined by CG and SNP array on multiple passages Supplementary Data 7 Overview of plasmid insertion sites Supplementary Data 8 Sanger sequencing validation of plasmid insertion sites Supplementary Data 9 R code used for gene level differential expression analysis Supplementary Data 10 R code used for exon level differential expression analysis
- Record: found
- Abstract: found
- Article: not found
HEK293 cell line: a vehicle for the expression of recombinant proteins.
Philip Thomas, Trevor Smart (2005)
The HEK cell line has been extensively used as an expression tool for recombinant proteins since it was generated over 25 years ago. Although of epithelial origin, its biochemical machinery is capable of carrying out most of the post-translational folding and processing required to generate functional, mature protein from a wide spectrum of both mammalian and non-mammalian nucleic acids. Though popular as a transient expression system, this cell type has also seen wide use in stably transfected forms (i.e. transformed cells) to study a variety of cell-biological questions in neurobiology. The principal attributes which have made the HEK cell a popular choice among electrophysiologists to study isolated receptor channels include; its quick and easy reproduction and maintenance; amenability to transfection using a wide variety of methods; high efficiency of transfection and protein production; faithful translation and processing of proteins; and small cell size with minimal processes appropriate for voltage-clamp experimentation. These, and other attributes, also mean that complementary biochemical/cell biological evaluations of expressed proteins can be performed in concert with functional analyses to establish detailed pharmacological and biophysical profiles for the action of new drugs and their targets. The increased amount of sequence information available from the human genome has placed greater emphasis upon heterologous cell expression systems as targets for high throughput structure-function evaluation of novel drug targets and disease markers. Here we have highlighted some of the innate characteristics of the HEK cell in order that its suitability as a vehicle for the expression of a gene product can be assessed for particular needs. We have also detailed some of the standard methods used for transfection and obtaining functional data from electrophysiological recording techniques.
Author and article information
Contributors
Meng How Tan:
ORCID: http://orcid.org/0000-0003-3627-5586
mh.tan@ntu.edu.sg
Say Kong Ng:
ORCID: http://orcid.org/0000-0002-4196-2754
ng_say_kong@bti.a-star.edu.sg
Journal
Journal ID (nlm-ta): Sci Rep
Journal ID (iso-abbrev): Sci Rep
Title:
Scientific Reports
Publisher:
Nature Publishing Group UK
(London
)
ISSN
(Electronic):
2045-2322
Publication date
(Electronic):
14
November
2019
Publication date PMC-release: 14
November
2019
Publication date Collection: 2019
Volume: 9
Electronic Location Identifier: 16768
Affiliations
[1
]ISNI 0000 0004 0485 9218, GRID grid.452198.3, Bioprocessing Technology Institute, Agency for Science, Technology and Research (A*STAR),
; Singapore, Singapore
Author information
Matthew Choo
http://orcid.org/0000-0002-1376-3352
Meng How Tan
http://orcid.org/0000-0003-3627-5586
Say Kong Ng
http://orcid.org/0000-0002-4196-2754
Article
Publisher ID: 53391DOI: 10.1038/s41598-019-53391-z
PMC ID: 6856173
PubMed ID: 31727983
SO-VID: 1c7ba1b6-371d-4cdb-a931-ab9e528bdc34
Copyright © © The Author(s) 2019
License:
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
History
Date
received
: 17
January
2019
Date
accepted
: 31
October
2019
Funding
Funded by: FundRef https://doi.org/10.13039/501100001381, National Research Foundation Singapore (National Research Foundation-Prime Minister's
office, Republic of Singapore);
Award ID: NRF2013-THE001-093
Award ID: NRF2013-THE001-093
Award ID: NRF2013-THE001-093
Award ID: NRF2013-THE001-093
Award Recipient
:
Andy Hee-Meng Tan
Terry Nguyen-Khuong
Meng How Tan
Say Kong Ng
Funded by: National Research Foundation Singapore (National Research Foundation-Prime Minister's
office, Republic of Singapore)
Funded by: FundRef https://doi.org/10.13039/501100001348, Agency for Science, Technology and Research (A*STAR);
Award ID: GlycoSing
Award ID: HighGlycoART
Award Recipient
:
Terry Nguyen-Khuong
Funded by: National Research Foundation Singapore (National Research Foundation-Prime Minister's
office, Republic of Singapore)
Funded by: National Research Foundation Singapore (National Research Foundation-Prime Minister's
office, Republic of Singapore)
Categories
Subject:
Article
Custom metadata
issue-copyright-statement © The Author(s) 2019
ScienceOpen disciplines: Uncategorized